arXiv subjects

Raphaëlle Chaine

Publications and source records attributed to Raphaëlle Chaine.

5 recordsLinked to original sources

arXiv2021

Arbitrary order principal directions and how to compute them

Curvature principal directions on geometric surfaces are a ubiquitous concept of Geometry Processing techniques. However they only account for order 2 differential quantities, oblivious of higher order differential behaviors. In this paper, we extend the concept of principal directions to higher orders for surfaces in R^3 by considering symmetric differential tensors. We further show how they can be explicitly approximated on point set surfaces and that they convey valuable geometric information, that can help the analysis of 3D surfaces.

cs.CG↗

arXiv2021

Baseline Skinning for Point Sets of Articulated Bodies

General skinning techniques aim to deform the surface of an articulated model following the pose change of a skeleton. Their rapidity makes them ideal tools for real-time animation purposes. However, popular skinning algorithms are simple, but they tend to generate undesirable geometric artefacts. In our work, we consider skeletons given in the form of sphere-mesh models controlling both the pose and morphology of the shape that is either described as a mesh or a raw point set. We propose a novel skinning method that encodes the point set details above a bundle of baselines covering the sphere-mesh. In particular, we propose a geometrical model of the baseline and detail direction evolution during bone twisting and joints bending rotations. Our approach works directly on point sets and thus preserves the accuracy of the initial sampling. It further avoids computing a weight per point or a costly explicit muscle modelling step. We evaluate our method on several articulated body point sets, showing that it creates fewer artefacts than classical methods.

cs.CG↗

arXiv2020

FAKIR: An algorithm for revealing the anatomy and pose of statues from raw point sets

3D acquisition of archaeological artefacts has become an essential part of cultural heritage research for preservation or restoration purpose. Statues, in particular, have been at the center of many projects. In this paper, we introduce a way to improve the understanding of acquired statues representing real or imaginary creatures by registering a simple and pliable articulated model to the raw point set data. Our approach performs a Forward And bacKward Iterative Registration (FAKIR) which proceeds joint by joint, needing only a few iterations to converge. We are thus able to detect the pose and elementary anatomy of sculptures, with possibly non realistic body proportions. By adapting our simple skeleton, our method can work on animals and imaginary creatures.

cs.GR↗

arXiv2017

Sparse Geometric Representation Through Local Shape Probing

We propose a new shape analysis approach based on the non-local analysis of local shape variations. Our method relies on a novel description of shape variations, called Local Probing Field (LPF), which describes how a local probing operator transforms a pattern onto the shape. By carefully optimizing the position and orientation of each descriptor, we are able to capture shape similarities and gather them into a geometrically relevant dictionary over which the shape decomposes sparsely. This new representation permits to handle shapes with mixed intrinsic dimensionality (e.g. shapes containing both surfaces and curves) and to encode various shape features such as boundaries. Our shape representation has several potential applications; here we demonstrate its efficiency for shape resampling and point set denoising for both synthetic and real data.

cs.CG↗

arXiv2013

Practical Reduction of Edge Flip Sequences in Two-Dimensional Triangulations

The development of laser scanning techniques has popularized the representation of 3D shapes by triangular meshes with a large number of vertices. Compression techniques dedicated to such meshes have emerged, which exploit the idea that the connectivity of a dense mesh does not deviate much from the connectivity that can be constructed automatically from the vertex positions (while possibly being guided by additional codes). The edge flip is one of the tools that can encode the differences between two meshes, and it is important to control the length of a sequence of flips that transform one triangulation into another. This paper provides a practical solution to this problem. Indeed, the problem of determining a minimal sequence of edge flips between two triangulations is NP-complete for some types of triangulations including manifold triangulations of surfaces, so that it is necessary to develop heuristics. Moreover, it is sometimes difficult to establish a first sequence of flips between two meshes, and we advocate a solution based on the reduction of an existing sequence. The new approach we propose is founded on the assignment of labels to identify the edges, with a property of label transfer during a flip. This gives a meaning to the tracking of an edge in a sequence of flips and offers the exploitation of very simple combinatorial properties. All the operations are performed directly on the sequence of labels denoting the edges to be flipped, almost regardless of the underlying surface, since only local temporary connectivity is involved.

cs.CG↗